Increase in incidence of disease due to diagnostic drift: primary liver cancer in Denmark, 1943-85.

OBJECTIVE: To examine the extent to which changes in diagnostic methods and classification are responsible for the striking increase in incidence of primary liver cancer in Denmark since 1943.
DESIGN: Analysis of the time trends in sex specific, age standardised incidence of primary liver cancer and unspecified liver cancer (either secondary without known primary cancer or not specified as primary cancer) in the entire population from 1943 to 1985. By review of the 727 notifications from three periods of 5 years (1948-52, 1963-7, and 1978-82) the changes in histological diagnosis and classification were assessed.
SETTING: Denmark.
SUBJECTS: Notifications of liver cancer to the Danish cancer registry.
RESULTS: Concomitant with the increase in primary liver cancer, the incidence of the unspecified liver cancer declined. The proportion of histologically diagnosed primary liver cancer rose from 85% to 98%, whereas the proportion for unspecified liver cancer rose from 12% to 51%. When the proportion of primary versus unspecified liver cancer obtained by histological diagnosis was extrapolated to all cases, the annual incidence of primary liver cancer was 4.4 rather than 1.6 per 100,000 population in 1948-52 and 6.0 rather than 5.5 per 100,000 in 1978-82.
CONCLUSION: The increase in the incidence of primary liver cancer may be much smaller than the numbers of registered cases indicate. This example emphasises the need to consider diagnostic drift in time trend studies of disease incidence.